Log body perimeters

Sunday morning. You take the tape measure out of the drawer for the weekly perimeter check. Waist, chest, arms, hips — four numbers in cm. You tell the agent the whole set in one sentence and move on.

What the agent needs to hear

Behind your sentence, the agent works backwards to two pieces: which perimeters and the number for each. The site labels are the standard ones — waist, chest, biceps (or arm), thigh, hip, neck, calf, forearm. Bilateral sites can carry a side qualifier (« left biceps 38, right 37.5 ») or a single value the agent records under both.

Date defaults to today. Unit defaults to cm (or inches if your profile is imperial). « Waist 86, chest 110, arm 38 » carries three sites in one sentence — the agent splits them and shows you a preview card with each one ready to save.

All-at-once vs single-perimeter

Most users log a set at once on a recurring day. The all-at-once shape — « waist 86, chest 110, arm 38, hip 96 » — is the fastest path. The agent reads each label-number pair, validates that the labels are recognized sites, and saves the whole set as one weekly entry.

Single-perimeter logs work the same way. « Waist 86 today » is fine when that’s all you measured. The trend reader stitches partial logs across days into a continuous series per site, so you don’t need to log every perimeter every time for the trend to read cleanly.

What this signal is for

Perimeter logs feed two downstream consumers. The per-site trend chart shows how each measurement is moving over weeks (waist down during a cut, arms holding during a recomp, hips up during a bulk — the curves read independently). The longitudinal narrative combines perimeters with weight and training data to read overall body change as one story.

Consistency is what makes the trend honest. Same tape, same anchor points (« measure the waist at the navel, not the smallest part »), same time of day, same state (fasted morning, not after a meal). The agent doesn’t audit your protocol — it logs what you say — but a series that bounces around because the measuring conditions varied won’t tell you anything real about the underlying change.

When the agent gets it wrong

If a site label resolved wrong (you said « arm » meaning biceps, the agent recorded forearm), name the site: « that was biceps, not forearm ». If two sites swapped values on the card (waist 86 and chest 110 inverted), call the assignment: « 86 is waist, 110 is chest ».

If a value got read in the wrong unit (« 38 » got logged as inches in a cm profile), correct it: « 38 cm, not inches ». And if you measured a site you don’t have a label for in your usual set (« measured my neck for the first time, 38 cm »), the agent will accept it and add neck to your tracked sites going forward.

What makes the log worth keeping

Three things decide whether this perimeter log helps you later: the site labels match where you actually measured (a chest reading logged as biceps corrupts both site trends until corrected), the measurement conditions match what you can repeat (anchor points, time of day, tape tension), and the values are what the tape actually read (no rounding down to the goal number — the trend chart needs honest inputs to draw the real curve). The system reads perimeters as the slower-moving counterpart to weight; a misattributed or padded reading shows up as a phantom shift in the trend. Measure the same way each time, say the numbers, move on.